Behavior-Based Detection of Abnormal Power Consumption for Power Saving

Abstract

Abnormalities are caused by incorrect or inappropriate behaviors or appliance malfunctions. They may lead to electricity waste and safety hazards. This paper describes a novel appliance management system for detecting abnormal power consumption in convenience stores based on power meters. Our system detects abnormal power consumption through historical behavior models. Generalized extreme studentized deviate (GESD) and regression methods are applied to build behavior models. The behavior based abnormal detection methods can assist in preventing these waste and safety problems and improve the appliance management to achieve power saving.